Effect of m1A regulators on prognosis of colorectal cancer

Abstract

Background Colorectal cancer (CRC) is one of the most serious public health problems. N1-methyladenosine modification appears to play a significant role in colorectal cancer development. Herein, we attempted to develop a prognostic prediction model to predict colorectal cancer prognosis using multiple m1A regulators and clinical characteristics. Methods The TCGA database was used to evaluate the expression of the m1A gene in CRC, and clustering analysis was carried out. The prognostic model of CRC was created using the Limma software, K-M survival analysis, and multivariate Cox regression, and it was then verified using the GEO database. Results We comprehensively evaluated m1A modification patterns and identified m1A subtypes used clustering analysis in CRC. Limma package was used to identify 17 differentially expressed m1A regulators in CRC patients, including 14 up-regulated regulators and 3 down-regulated regulators. K-M survival analysis identified three m1A regulators (TRMT61B, HNRNPM, and YTHDC1) associated with overall survival in CRC patients. A gene signature based on these three m1A regulators was developed using multivariate Cox regression which was efficient in predicting long-term prognosis of CRC patients. In addition, multivariate Cox regression analysis demonstrated that risk score (HR: 2.598, 95% CI: 1.226–5.506, P = 0.013) and TNM stage (HR: 1.923, 95% CI: 1.235–2.993, P = 0.004) are two independent prognostic factors. Next, we constructed a nomogram with a concordance index of 0.720 based on gene signature and TNM stage to provide a personalized overall survival prediction in CRC patients. Compared with TNM stage, the nomogram exhibited excellent performance in predicting prognosis. The AUC of 1-, 3- and 5-year OS rates of TNM stage were 0.720, 0.745 and 0.742; whereas the AUC of 1-, 3- and 5-year OS rates of nomogram were 0.721, 0.760 and 0.772 in TCGA database, respectively. Last but not least, the expression of three m1A regulators were verified by q-PCR experiment and the prognostic performance of gene signature and nomogram were validated in a cohort of GEO datasets. Conclusion We have constructed and verified a novel prognostic gene signature and a nomogram based on m1A regulators that might effectively promote overall survival prediction in CRC patients.